Recent Projects

Project: Advanced Research of (Large Language Model)LLM accelerating IC design 2023.5-present

• Project Role: Technical Expert
• Project Responsibilities:
  • Fine-tune pre-trained LLMs on Verilog datasets
  • Construct an evaluation framework for functional analysis
  • Conduct distributed training
• Project Description: Automating IC design could obviate a significant amount of human error from the engineering process and lead to fewer errors. Verilog is a popular hardware description language to model and design digital systems, thus generating Verilog code is a critical first step. In this project, we use emerging large language models (LLMs) to write high-quality verilog code.

Project: Advanced Research of Application Processor for In-Vehicle Infotainment (IVI) and Advanced Driver Assistance Systems 2021.9-2023.6

• Role: Area Product Owner(APO) of DSP-Subsystem
• Project Responsibilities:
  • ADAS and Audio algorithm design for IVI system;
  • User requirement analysis and competitive product research;
  • IP(DSP) vendor analysis, evaluation and selection;
  • Algorithm vendor evaluation.

Project: Ultra-low power edge AI chip 2019.9-2021.9

• Project Role: Algorithm Development
• Project Responsibilities:
  • Binary Neural Networks development;
  • Group convolutional neural network development;
  • Hardware design supporting.
• Project Description: The Ultra-low Power Edge AI Chip can support ultra-long standby and operation time of portable intelligent devices, e.g. smart watches, earphones, mobile phones, laptops, surveillance cameras, etc., with power consumption as low as 6 milliwatts. This chip won 2021 HK ICT Startup (Hardware and Devices) Silver Award.

Project: Feasibility Investigation of Deep Learning in Device Modeling (ARD/226) 2018.6-2019.9

• Project Role: Core Algorithm Development
• Project Responsibilities:
  • Develop deep learning application for advanced device modeling;
  • HK Innovation and Technology Fund application.
• Project Description: Device model plays important role in integrated circuit (IC) design and good device model is the prerequisite to ensure the success of design. This project targets at developing deep learning based device modeling methodology. The model will be implemented in a circuit simulator for verification. The core algorithm developed in this project can be further applied to applications such as radio frequency (RF) and power device.